Hybrid evaluation of expressions in DBMS

• US 7,895,187 B2
• Filed: 12/21/2006
• Issued: 02/22/2011
• Est. Priority Date: 12/21/2006
• Status: Active Grant

• Current Assignee: iAnywhere Solutions Incorporated (SAP SE)
• Original Assignee: Sybase Incorporated (SAP SE)
• Inventors: Bowman, Ivan Thomas
• Primary Examiner(s): Robinson; Greta L
• Assistant Examiner(s): Chang; Jeffrey
• Application Number:
  US11/642,622
• Publication Number:
  US 20080154818A1
• Time in Patent Office: 1,524 Days
• Field of Search: 707/2, 707/713
• US Class Current: 707/713
• CPC Class Codes: G06F 16/2454 Optimisation of common expressions
• CPC Class Codes: G06F 16/24542 Plan optimisation

Abstract

A hybrid approach for evaluating expressions in database management systems is described herein. For each expression in an execution location (XL), an execution strategy is selected. The strategies considered are lazy, lazy with memoization, eager, and eager with memoization. The execution strategies are assigned in a way that minimizes the estimated cost of evaluation.

Claims

1. A method of evaluating an expression in a database management system, comprising:

(a) determining whether said expression is strictly needed, wherein said determining step comprises determining if one or more arguments of said expression must be evaluated before said expression itself can be evaluated;

(b) determining a number of parents of said expression;

(c) evaluating said expression at an execution location using every argument in said expression, when it is determined that said expression has multiple parents, when said expression is strictly needed; and

(d) evaluating said expression at most once at an execution location using a subset of arguments in said expression, regardless of whether an evaluation of said expression is needed more than once at said execution location, when it is determined that said expression has a single parent.

2. The method of claim 1, wherein step (a) comprises:

determining that said expression is strictly needed if one or more of (i) and (ii) is true; and

determining that said expression is not strictly needed if (i) and (ii) are false;

wherein (i) and (ii) are defined as follows;

(i) said expression is strictly needed by an execution location;

(ii) said expression is reachable by one or more strict edges from a strictly needed expression.

3. The method of claim 1, further comprising:

determining a first estimated cost of evaluating said expression eagerly;

determining a second estimated cost of evaluating said expression lazily;

evaluating said expression eagerly if said first estimated cost is less than said second estimated cost; and

evaluating said expression lazily if said second estimate cost is less than said first estimated cost.

4. The method of claim 3, wherein said second estimated cost comprises an estimated cost of evaluating said expression, and an estimated administrative cost associated with valid flags and back links.

5. The method of claim 1, wherein said expression is a scalar expression.

6. The method of claim 1, wherein said subset of arguments of said expression are used for evaluating said expression, when it is determined that said expression is true if any one of the arguments in said subset are also true.

7. An article of manufacture including a computer-readable medium having instructions stored thereon that, when executed by a processing device, cause the processing device to perform operations comprising:

determining whether said expression is strictly needed wherein said determining step comprises determining if one or more arguments of said expression must be evaluated before said expression itself can be evaluated;

determining a number of parents of said expression;

evaluating said expression at an execution location using every argument in said expression, when it is determined that said expression has multiple parents, and said expression is strictly needed; and

evaluating said expression at most once at an execution location using a subset of arguments in said expression, regardless of whether an evaluation of said expression is needed more than once at said execution location, when it is determined that said expression has a single parent.

8. The article of manufacture of claim 7, said operations further comprising

determining that said expression is strictly needed if one or more of (i) and (ii) is true;

and

determining that said expression is not strictly needed if (i) and (ii) are false;

wherein (i) and (ii) are defined as follows;

said expression is strictly needed by an execution location;

(ii) said expression is reachable by one or more strict edges from a strictly needed expression.

9. The article of manufacture of claim 7, said operations further comprising:

determining a first estimated cost of evaluating said expression eagerly;

determining a second estimated cost of evaluating said expression lazily;

evaluating said expression eagerly if said first estimated cost is less than said second estimated cost; and

evaluating said expression lazily if said second estimate cost is less than said first estimated cost.

10. The article of manufacture of claim 9, wherein said second estimated cost comprises an estimated cost of evaluating said expression, and an estimated administrative cost associated with valid flags and back links.

11. The article of manufacture of claim 7, wherein said expression is a scalar expression.

12. A compiler that evaluates an expression in a database management system, comprising:

a first module configured to determine whether said expression is strictly needed, and configured to determine if one or more arguments of said expression must be evaluated before said expression itself can be evaluated;

a second module configured to determine a number of parents of said expression

a third module configured to evaluate said expression at an execution location using every argument in said expression, when it is determined that said expression has multiple parents, and said expression is strictly needed; and

a fourth module configured to evaluate said expression at most once at an execution location using a subset of arguments in said expression, regardless of whether an evaluation of said expression is needed more than once at said execution location, when it is determined that said expression has a single parent.

13. The compiler of claim 12, further comprising:

a fifth module configured to determine that said expression is strictly needed if one or more of (i) and (ii) is true; and

a sixth module configured to determine that said expression is not strictly needed if (i) and (ii) are false;

wherein (i) and (ii) are defined as follows;

(i) said expression is strictly needed by an execution location;

(ii) said expression is reachable by one or more strict edges from a strictly needed expression.

14. The compiler of claim 12, further comprising:

a fifth module configured to determine a first estimated cost of evaluating said expression eagerly;

a sixth module configured to determine a second estimated cost of evaluating said expression lazily;

a seventh module configured to evaluate said expression eagerly if said first estimated cost is less than said second estimated cost; and

an eighth module configured to evaluate said expression lazily if said second estimate cost is less than said first estimated cost.

15. The compiler of claim 14, wherein said second estimated cost comprises an estimated cost of evaluating said expression, and an estimated administrative cost associated with valid flags and back links.

16. The compiler of claim 12, wherein said expression is a scalar expression.